On 4 April 2011 16:27, Jim Mulder <[email protected]> wrote: > On the other hand, for problems which occur intermittently in > production environments and are not reproducible in the development > environment, how do the debugging facilities available on Windows, > Java and Unix in general compare to those facilities on z/OS?
Your point is good, but in my experience it's a matter of customer expectations. Our products run on and interconnect z/OS and other platforms. When there is a failure on z/OS, the customer's support group typically sends us a dump and wants to know what happened. Telling them we will "try to reproduce it in the lab" doesn't cut it; they expect us to be able to diagnose and fix the problem based on the available evidence, and to fail to do so is the exception for them and us. Reproducing the problem is only to confirm that the diagnosis and fix are correct. If we get a similar failure on Windows at the same customer site, they have no expectation that we will be able to resolve anything based on the report. Indeed they rarely send any supporting evidence beyond a description of the bad behaviour, and perhaps a log file. Beyond that, the expectation is that we will supply further instructions to turn on debug/trace mode (that is not normally enabled in production for performance reasons) and try to reproduce the problem. Even though platforms like Windows, UNIX, and Java have dumps, they are rarely used, and the tools and techniques necessary to use them are not widely understood. Perhaps naturally, the attitudes of our own support people and developers follow to some extent. These attitudes clash when one of our components that runs on z/OS UNIX is in the picture. With my MVS roots, I often arrange to take a SLIP dump, but then I am the one who has to turn that into C line numbers and variable contents for someone to look at and analyse, and use to try to reproduce the problem. As customers move work from z/OS, inevitably the support expectations follow, and what should be a selling point for z/OS loses its effectiveness. Tony H. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
